Rebuild without rust dev channel

[conda-forge-admin]

Closes #141
Hi! This is the friendly automated conda-forge-webservice.

I've started rerendering the recipe as instructed in #141.

If I find any needed changes to the recipe, I'll push them to this PR shortly. Thank you for waiting!

Here's a checklist to do before merging.

• Bump the build number if needed.

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[github-actions]

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

This message was generated by GitHub actions workflow run https://github.com/conda-forge/ruff-feedstock/actions/runs/6447769194.

[hmaarrfk]

@conda-forge-admin please rerender

[charliermarsh]

I can't exactly remember why we moved to the dev channel -- I think we upgraded to a version of Rust that somehow wasn't available in the main channel?

[hmaarrfk]

right. I feel like that might have been the case. might as well revert things right?

[charliermarsh]

Yeah, as long as we don't run into the same problem again in the future. E.g., is v1.73.0 up yet on the non-dev channel? I think so, right? It's this: conda-forge/rust-feedstock#168?

[hmaarrfk]

is ruff really on bleeding edge rust all the time? typically that trend seems to die down eventually.

1.73 is on the main channel as can be seen in the files list:
https://anaconda.org/conda-forge/rust/files

[hmaarrfk]

I'm happy to leave the file in there, with a comment explaining when to to use the dev channel, but I don't think we should be leaving it on all the time.

[charliermarsh]

is ruff really on bleeding edge rust all the time?

Honestly, kind of... We likely would've already upgraded to 1.73 if it were supported on Homebrew. The limiting factor is typically support for new versions in the various registries, since we've been burned in the past by overly aggressive early upgrades.

[hmaarrfk]

I'm somewhat confused:

We likely would've already upgraded to 1.73 if it were supported on Homebrew.

You claim you want to be bleeding edge.

since we've been burned in the past by overly aggressive early upgrades.

You mention you don't want aggressive updates? but yet you want to to use bleeding edge rust?

The thing is that conda-forge reach has gone beyond the "early adopters" where it was OK to be on the bleeding edge and breaking on daily basis.

CFEP-05 speaks to this in some words
https://github.com/conda-forge/cfep/blob/main/cfep-05.md?plain=1#L95

users of conda-forge expect things to be working, maybe a version old or two, while bugs in the bleeding edge are worked out.

I do not thing that we want to be "always running on bleeding edge" unless we absolutely have to (see the exception in CFEP-05)

The maintenance churn, of using "development branches or development labels" should act as a disincentive to using the dev channel.

Because today, rust 1.73 is available on conda-forge wide, I do not see a point of using rust 1.74 pre-release.
If you insist on using prerelease rust, then this channel should also publish to a pre-release channel, limiting the adoption to the wider audiance.

[zanieb]

we've been burned in the past by overly aggressive early upgrades

I'm sure Charlie can respond to the rest but just to clarify this point — we've been burned because the downstream packagers (i.e. HomeBrew / Conda) do not support the new release before we upgrade. We're not getting burned by bugs in the new Rust releases.

[hmaarrfk]

Thank you @zanieb

We're not getting burned by bugs in the new Rust releases.

Understood.

we've been burned because the downstream packagers (i.e. HomeBrew / Conda) do not support the new release before we upgrade

conda's default channel is often behind quite a few versions. I can only imagine that they (Anaconda/Binstar) have different priorities.

I really do appreciate having upstream's ears for package maintenance, I think it can greatly help expedite things.

I can only speak for as a member of the conda-forge core team but:

Your help ensuring our version of rust stays up to date would be greatly help ensure a timely release of the latest rust packages. Generally speaking, this has been how my personal involvement grew.

Given the new information, I will amending to my previous comment

Because today, rust 1.73 is available on conda-forge wide, I do not see a point of using rust 1.74 pre-release.
If you insist on using prerelease rust, then this channel should also publish to a pre-release channel, limiting the adoption to the wider audiance. If the rust version is lacking, a PR to the rust-feedstock to update it to the latest release version would help the community and be the preferred method of "updating" the package. I would be glad to review the rust-feedstock PR, but it seems that it already has the eyes of many active contributors to conda-forge including a few core members.

[charliermarsh]

Thanks @zanieb, and sorry for the confusion @hmaarrfk. To clarify my initial comment: we like to be on the latest stable Rust, and we tend to upgrade as soon as we can. In the past, there was at least one release for which we upgraded "too fast", and broke our conda-forge and/or Homebrew builds, which is likely why we ended up on the dev channel here. Since then, we've deferred upgrading our Rust version until it's available on the stable channel.

So: do like to be on latest stable as soon as we can (which led to me confusingly saying "yes" when asked "is ruff really on bleeding edge rust all the time?"), but we've also stopped upgrading before conda-forge, which means this change should be totally fine.

[charliermarsh]

Feel free to merge :)